FeII4L4 Tetrahedron Binds to Nonpaired DNA Bases.

A water-soluble self-assembled supramolecular FeII4L4 tetrahedron binds to single stranded DNA, mismatched DNA base pairs, and three-way DNA junctions. Binding of the coordination cage quenches fluorescent labels on the DNA strand, which provides an optical means to detect the interaction and allows the position of the binding site to be gauged with respect to the fluorescent label. Utilizing the quenching and binding properties of the coordination cage, we developed a simple and rapid detection method based on fluorescence quenching to detect unpaired bases in double-stranded DNA.

Interactions between DNA and synthetic molecules have enabled numerous applications in biomedicine, gene regulation, and disease diagnosis.[1] For instance, many DNA-targeting anticancer drugs have been designed and developed based on their binding to double-stranded DNA (dsDNA).[2] Besides the canonical B-form antiparallel DNA double helix, there are several other structural forms of DNA, including parallel duplexes, triplexes, and quadruplexes, and three- and four-way DNA junctions. These structures occurring at specific genomic locations have various biological functions in natural systems,[3] but have also been used to build DNA nanostructures and devices.[4]

The three-way junction (3WJ) is an uncommon DNA structure, although it is implicated in many DNA metabolic processes, such as replication, transcription, recombination, and repair.[5] Errors in these processes can lead to DNA mutations, which may cause a cascade of problems in gene expression.[6] Hence, the development of recognition probes for 3WJs and DNA duplexes that contain a small number of unpaired bases may lead to the emergence of more efficient diagnostic tools.[7]

Coordination driven self-assembly has enabled the generation of useful three-dimensional molecules[8] and materials[9] with tunable structures and properties. Numerous applications of these metallosupramolecular complexes in nucleic acid research have been developed in recent years.[10] Hannon, Coll, and co-workers have employed metallosupramolecular cylinders to induce the formation of three-way DNA and RNA junctions.[11] The binding of plasmid DNA and G-quadruplexes by coordination complexes,[12] including two M4L6 tetrahedral cages,[12c,12e] has also been reported, which demonstrates the potential for application of the metallosupramolecular cages in recognition of DNA structures. Building upon these pioneering studies, here we report the use of a simple FeII4L4 tetrahedron (1, Figure a)[13] as a site-selective binder for 3WJs and base pair mismatches. The binding of 1 quenches the fluorescence of a proximate fluorophore, enabling an all-optical readout for the sensing of these DNA structures.

Figure 1

Fluorescence study of the interaction between cage 1 and different types of DNA. (a) Molecular structure of cage 1; the SO42– counterions are omitted for clarity. (b) Effects of 1 (0.2 μM) on the fluorescence intensity of different DNA structures (0.1 μM) labeled with the FAM fluorophore. (c) Quenching efficiency based on the concentration ratio of cage 1 to the whole DNA structures (0.1 μM). Inset: Kd values for different DNA structures. Further details on data fitting are given in Figure S4. All fluorescence measurements were performed in PBM buffer (10 mM phosphate buffer, 10 mM MgSO4, pH 7.5).

We selected tetrahedral cage 1 as a suitable candidate for DNA binding due to its water solubility and stability under biologically relevant conditions (Supporting Information sections S1 and S2). We also hypothesized that the trigonal three-dimensional shape of the tetrahedral cage may contribute to the recognition of 3WJ, as in the case of the 3-fold-symmetric cylinders reported by Hannon.[11] Cage 1 was found by fluorescence spectroscopy to quench the fluorescent dye 6-carboxyfluorescein (FAM) attached to different DNA strands, showing a preference for single-stranded DNA (ssDNA) and 3WJ DNA. Studies of the interaction between 1 and 3WJ revealed 1 to bind preferentially at the central cavity of a 3WJ, where base-pairings are loose or even absent, according to recent reports.[14] The inference that 1 bound preferentially to unpaired sites thus led to the detection of a series of base mismatches in DNA double strands.

Bulk fluorescence measurements were used to study the interaction of cage 1 (Figure a) with different DNA structures (Supporting Information section S3). We found the cage to quench the fluorescence emissions of FAM dye molecules covalently bound to the 5′ end of DNA strands, but to have no effect on the fluorescence of the dye itself when free in solution (Figure S10a,b). Cage 1 was treated with four different kinds of DNA structures, each labeled with FAM (Figure b): ssDNA, dsDNA, 3WJ DNA, and four-way DNA junction (4WJ) DNA. The fluorescence spectrum of each sample was measured before and after mixing with cage 1 (Figure b). We observed cage 1 to quench the FAM fluorescence in the samples containing ssDNA and 3WJ by 69% and 75%, respectively. Thus, 1 shows higher affinities for these structures, as compared to dsDNA and 4WJ, for which only 40% and 38% of quenching were observed. The quenching efficiency (QE, Supporting Information section S3) is expected to depend on the relative concentrations of 1 and DNA. Figure c shows the dependence of the QE on the ratio of cage 1 to DNA for four different DNA structures. Dissociation constants (Kd) were determined for the different DNA structures (Figure c and Supporting Information section S3, Table S3), indicating stronger interactions of cage 1 with 3WJ and single-stranded DNA. Moreover, we found only the fully assembled cage to cause fluorescence quenching; addition of either FeII or the precursor aldehyde and triamine subcomponents of 1 to the 3WJ had a minimal effect on fluorescence intensity (Figure S10c).

To gauge further the binding between cage 1 and 3WJ, gel electrophoresis was applied to DNA samples in the presence (+) and absence (−) of cage 1. Gel results (Figure a, Supporting Information section S5) showed a clear change of the 3WJ band (red square in Figure a) following the addition of cage 1. A slight shift of the ssDNA band was also observed. In contrast, no position difference was observed for dsDNA. Although the band of the 4WJ became broadened in the presence of 1, the position of the main band did not change. The addition of increasing amounts of cage 1 into 3WJ led to a migration of the band (Figure b). These gel results confirm the specific binding between cage 1 and 3WJ, which increased the molecular weight of the complex, partly neutralized the DNA charge, and in consequence slowed the mobility of 3WJ. Fluorescent melting experiments were also performed to investigate the binding in Supporting Information section S4.

Figure 2

Gel results: (a) polyacrylamide gel electrophoresis of studied DNA structures (5 μM) with (+) and without (−) cage 1 (50 μM). The highlighted band has been investigated further. (b) Gel shift of 3WJ (1 μM) upon titration with increasing concentrations of cage 1. The same ladder is used in panel b).

Varying the distance between the branch point and the FAM label on different 3WJ (Figure a) allowed for the binding mode of 1 to be probed. Previous reports[11,15] led us to the hypothesis that cage 1 binds to the 3WJ at the central branch point. Increasing the distance between the central cage binding site and the fluorescent label (FAM) is thus expected to reduce fluorescence quenching. The results of this experiment were in agreement with our hypothesis (Figure a); an apparent linear relationship between the normalized fluorescence intensity and the distance was observed, agreeing with the presence of cage 1 at the branch point of the 3WJ.

Figure 3

Experiments to determine the location of binding. (a) Effect of the distance (x base pairs, bp) between the label and the branch point of 3WJ on the fluorescence intensity of FAM in the presence of cage 1. Values were normalized against the mean fluorescence intensities of 3WJ-x (x = 8, 11, 14, 17, 20) in the absence of cage 1. Data were averaged over three experimental repeats. (b) Effect of mismatch (shown in red) on fluorescence quenching by cage 1 (0.2 μM) at the branchpoint of 4WJ (0.1 μM).

The complementary bases closest to the branchpoint of three-way DNA junctions are unstable and may not pair.[14] This flexible configuration may provide a central trigonal cavity for the triangular shaped cage to bind, similarly to previously reported triangular cylinders.[11] Cage 1 has a strong affinity toward ssDNA (Figure a), the looser structure of which is inferred to an optimal configuration for binding 1. In analogous fashion, we hypothesized that cage 1 binds more readily to the unpaired bases present in the central cavity of the 3WJ. To test this hypothesis, we introduced mismatched bases at the branch point of a 4WJ to destabilize the structure (Figure S7).[16] As shown in Figure a, cage 1 minimally quenches the fluorescent dye on the 4WJ. This may be a result of the rigid branch point of the 4WJ in the presence of Mg2+ reducing the propensity of the cage to bind.[17] However, when two mismatched bases were introduced at the central branch point of the 4WJ, fluorescence quenching was increased (Figure b). This observation indicates that the unpaired bases enable the cage to bind with higher affinity to the previously less accessible branch point of the 4WJ. It is worth noting that 3WJ is much less stable than 4WJ even if the central bases are fully paired,[16b] and thus its own special molecular conformation and central trigonal cavity may play more important roles in the binding in this case, which implies the cage may recognize different types of DNA structures. Further discussion of the quenching mechanism is given in Supporting Information section S7.

The DNA-cage binding that occurred specifically at unpaired bases inspired us to explore the potential of cage 1 to sense base-pair mismatches in dsDNA. We first investigated the single-base mismatch at different locations along dsDNA (M1–M3 in Figure ). These results showed a clear dependence between the quenching efficiency and the position of the mismatch relative to the FAM label. In the case of M1, with the FAM 5 bases away from the mismatch, the quenching was minimal. However, the quenching efficiency increased in the cases of M2 and M3, in which the mismatch was closer to the label. Besides the GG mismatched M2, the cage is also sensitive to the other types of mismatches such as GA and GT at the same site (Figure S8a). We inferred that a single base mismatch was detected by the cage, most strongly for mismatches closest to the fluorescent label.

Figure 4

Mismatch detection in DNA duplexes (0.1 μM) using cage 1 (0.2 μM). The bar graph shows normalized fluorescence emission intensities at 520 nm of different cage 1/DNA mixtures excited at 495 nm. Mismatches in the dsDNA are marked in red.

On the basis of the distance between the mismatch and the label, we expected the quenching efficiency to be greater for M3 than M2. One possible explanation for the divergent observation is that the first A:T base pair of M2 next to the mismatch site was destabilized, rendering its behavior similar to that of a sequence containing two mismatches. Indeed, when sequences containing two mismatches (M4–M6) were investigated, we found that the quenching efficiency was increased in proportion to the distance between the mismatch point and label.

Finally, we introduced progressively more mismatched bases into the sequence at a constant distance from the FAM label (Figure , M6–M8). The degree of fluorescence quenching was observed to increase with the number of unpaired bases. A DNA bulge induced by more unpaired bases[18] also enhanced the quenching by cage 1 (Figure S8b). We infer regions containing more unpaired bases to provide a larger cavity in dsDNA, ensuring enhanced binding to 1. Kd values for the interaction between cage 1 and mismatched and bulged DNA can be found in Figures S4, S5, and Table S4.

Overall, the above findings establish a simple and rapid new method to detect unpaired bases in dsDNA. Compared with the classic DNA mismatch-binding ligands, including metal complexes, small organic molecules, and simple metal ions, which have been reported recently,[19] tetrahedral cage 1 possesses unique advantages as a mismatched DNA probe. First, although it is not a mismatch-selective binder, in the manner of other metal complexes and ions[20]1 can sense a variety of base mismatches (Figure S8a). Second, in contrast to the luminescent metalloinsertors,[21] the ability of cage 1 to give rise to fluorescence quenching allows for base mismatch detection at a lower concentration. In addition, cage 1 combined with a specifically designed fluorescently labeled probe strand would enable selective detection of target structures or sequences. Finally, the use of subcomponent self-assembly to prepare three-dimensional supramolecules enables the tuning of probe structure and selectivity. The encapsulation of guest molecules inside coordination cages could enable these guests to be released in the vicinity of specific DNA regions.

In conclusion, the interaction between FeII4L4 tetrahedron and DNA has been probed for the first time, and applied to detect mismatches in DNA base pairs. Compared with previous DNA binders, the fluorescence quenching property of cage 1 enables straightforward optical detection and hence molecular sensing applications. Our approach relying on metallosupramolecular complexes enhances the flexibility and expandability for future designs. Given the significance of DNA mismatch detection in the diagnosis of genetic diseases and the value of three-way junctions in DNA metabolic processes, cage 1 adds a promising compound for fluorescence assays especially in nanobiotechnology and biomedicine. We foresee other such metal–organic cages formed using subcomponent self-assembly to enable other new applications in medical and biological sensing.